antigravity-skill-orchestrator-v2

Pass

Audited by Gen Agent Trust Hub on Jul 2, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructions direct the agent to fetch a master catalog of skills from a remote GitHub repository (https://raw.githubusercontent.com/sickn33/antigravity-awesome-skills/main/CATALOG.md) when local capabilities are insufficient. The agent is tasked with scanning this document to identify relevant skills for the user's request.
  • [COMMAND_EXECUTION]: The orchestration workflow authorizes the agent to execute terminal commands and perform file editing using tools like claude-code, cursor, and codex-cli for tasks that are deemed simple enough to solve without invoking specialized skills.
  • [INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection due to its handling of external and shared data.
  • Ingestion points: Data enters the agent's context through a remote markdown catalog file and records retrieved from the agent-memory-mcp tool.
  • Boundary markers: The skill instructions do not prescribe the use of delimiters or specific markers to isolate external data from the agent's instruction set.
  • Capability inventory: The agent has significant capabilities, including shell access and the ability to modify project files.
  • Sanitization: There is no evidence of content validation, escaping, or filtering for data retrieved from the remote catalog or the memory tool.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 2, 2026, 01:05 PM
Security Audit — agent-trust-hub — antigravity-skill-orchestrator-v2